246 



The Morphology of the Andrcecium of the Fumariace^e. 

 By J. T. RoTHROCK. 



My observations were made upon Dicentra spectaiilis, a good 

 representative of the natural order Fumariacece, and well adapted 

 for the present purpose. The flower in this order is dimerous. 

 There are two small and scale-like sepals, and four petals in two 

 distinct sets. Before each of the two exterior and larger petals 

 ■stands a phalanx made up of three dimorphous stamens, i. e. the mid- 

 dle anther of each phalanx is two-celled, or complete ; the lateral 

 ones are one-celled or, as it were, dimidiate. Now, the question is, 

 Are the stamens morphologically two, but each divided by chorisls or 

 deduplication into three ? Or are they normally four, one before each 

 petal, but those belonging before the inner petals divided in some 

 way each into two half-stamens, and these half-stamens so separated 

 as to be associated, one on each side, with the stamen before each 

 outer petal, and united with it diadelphously, so as to make two pha- 

 langes, each of a complete stamen in its proper position, and of a 

 half stamen on each side out of position ? 



The latter view was suggested by De Candolle, and afterwards more 

 fully developed by Lindley. If this be the correct view, we should ex- 

 pect (going back to an early stage in the development of the flower) 

 to find the one-celled stamens originating before the inner petals, 

 where they belong ; and also, as they answer to a single leaf, we 

 ■should look for a connection between their fibro-vascular bundles, and 

 for no connection with those which supply the middle stamen. The 

 ■theory of this view, moreover, supposes a deduplication or chorisis, 

 i. e. a fission of each of one set of stamens into two, and then a 

 cohesion, more or less complete, of the severed parts with the un- 

 altered stamens. The opposing and first-named view supposes only 

 a chorisis, without displacement, but into three instead of two mem- 

 bers. Those botanists, therefore, who reject the idea of chorisis, can- 

 not very consistently adopt the CandoUean explanation; for that, in 

 fact, assumes a chorisis no less than the other view. 



The view of a simple chorisis, dividing the stamens of a diandrous 

 .andrcecium each into three, has been adopted by Gray, on general 

 .morphological considerations, and afterwards by the late M. Payer, on 

 •organogenic observations. 



Upon this view, we should expect to find all three members of a 

 phalanx originating in connection, and in the position which they oc- 

 cupy in the developed flower, viz., before the outer petals, leaving 

 the space before the inner petals vacant. And as each phalanx 

 morphologically represents a single trilbliolate leaf, its fibro-vascular 

 bundles should originate in connection, or in proximity, and should be 

 *uch as belong to one leaf. It is the object of this communication to 



